Automatic validation for crowd simulation: Test suite for a pedestrian simulator based on different scenarios

Abstract

Evacuation simulation and especially the determination of evacuation times is a very complex task. Moreover the prognosis of place and time of critical bottlenecks within the building during the evacuation is critical due to complex building structures and the correct pedestrian behavior. Therefore, an extensive validation and calibration of the simulation algorithms is an indispensable requirement for every simulation tool. An automatic test suite for different scenarios will facilitate this task yielding in proven, automated and reproducible results. The microscopic pedestrian simulator tested in this paper is developed by our group. The tool can be used to guide the crowd evacuation and prepare respond plans for emergent situations as reference to city council and law enforcement agency. It is important that the simulation results reveal the true behavior of pedestrian; for certain precaution actions can be taken in order to guarantee the safety of the crowd. In this paper, we documented the performance of our simulator tested with all 14 scenarios proposed by the RiMEA (Richtlinie fur Mikroskopische Entfluchtungs-Analysen) guideline. The test results show that our simulator passes all the tests. Moreover, our pedestrian simulator constantly improves its performance by cooperating with construction companies and government departments running on-site tests with first-hand data. Now it covers even emergency scenarios such as fire/smoke and floods.